Open classroom systems frequently use peer tutoring techniques as a means of individualizing instruction. This study investigated the effects of three variables on tutor and tutee performance: (1) the achievement level of the tutor; (2) brief tutor training in reinforcement and corrective feedback procedures; and (3) tutor expectancy about tutee performance. One hundred and twelve first graders were randomly selected to serve as tutees. Results indicated that both tutors and tutees learned a significant number of words. This positive effect was more pronounced for low achievers than for high achievers. Training of tutors significantly increased the pretest to posttest gains for both tutors and tutees. Peer tutoring guidelines that can be generated from this research include: (1) all children in a class, regardless of achievement level, should be selected to serve in the tutoring role; (2) brief tutor training in basic reinforcement and corrective feedback procedures is essential to an effective peer-tutoring program; and (3) expectancies about tutees' performance may result in less biased teaching behavior by peer tutors than by adult tutors. (SK)